Storage Tiering Using Heat Weights for Conflicting Application Demands
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional tiered data storage systems face inefficiencies due to conflicting application performance requirements, overloading of faster tiers, and insufficient data placement, leading to resource conflicts and waste of storage capabilities.
Innovation Solution
A computer-implemented method that uses application data to predict performance requirements, maintains a persistent control repository, and adjusts tiering based on calculated heat weights to optimize data placement across storage tiers, balancing space allocation and performance demands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If frequently accessed data is placed in faster storage tiers, then access speed is improved, but storage cost increases and faster tiers become overloaded
Solution Approach 1:
The system dynamically adjusts data placement across storage tiers based on real-time access patterns and performance requirements. The tiering manager continuously monitors application demands and recalculates optimal data locations, transitioning from static to dynamic tiering to balance speed and cost efficiently
Solution Approach 2:
The system changes the parameter of data placement by calculating heat weights that combine access frequency with application performance requirements. This parameter transformation enables more nuanced data placement decisions that consider both speed needs and cost constraints simultaneously
2Productivity
If more data is placed in faster storage tiers to meet application performance requirements, then access performance is improved, but resource conflicts and tier overloading occur
Solution Approach 1:
The system implements feedback mechanisms where the tiering manager continuously monitors application performance requirements and access patterns. This feedback loop enables the system to adjust data placement dynamically, preventing tier overloading by redistributing data based on actual performance needs rather than static allocations
Solution Approach 2:
The system performs preliminary calculations of heat weights that incorporate both access frequency and application performance requirements before making data placement decisions. This preliminary assessment prevents unnecessary data movements and reduces system complexity by making informed placement decisions in advance
3Ease of operation
If traditional heat calculation methods are used for data tiering, then data placement is simplified, but application-specific performance requirements are not met
Solution Approach 1:
The system transforms the traditional heat calculation by introducing a new parameter that combines access frequency with application performance requirements. This parameter change maintains the simplicity of heat-based tiering while significantly improving placement precision to meet specific application demands
Solution Approach 2:
The modified heat weight calculation serves multiple functions simultaneously: it captures access patterns, evaluates application performance requirements, and determines optimal data placement. This multi-functionality maintains operational simplicity while achieving precise data placement
Data Source
AI summary
A computer-implemented method (CIM), according to one embodiment, includes obtaining application data associated with interaction of applications of host servers that use a data storage system and using the application data to predict performance requirements for the applications with respect to the data storage system. The method further includes maintaining a persistent control repository having parameter fields. A first and second of the parameter fields detail, for extents of data stored in tiers of the data storage system, associated ranges of acceptable performance targets, and a third and fourth of the fields detail, associated arrival times for requests from the applications for the extents of data. The method further includes using the persistent control repository to calculate heat weights for the extents of data, and causing a tiering manager to adjust a current tiering of the extents of data in the tiers based on the calculated weights.